Intravascular Optical Coherence Tomography Image Segmentation Based on Support Vector Machine Algorithm

Intravascular optical coherence tomography (IVOCT) is becoming more and more popular in clinical diagnosis of coronary atherosclerotic. However, reading IVOCT images is of large amount of work. This article describes a method based on image feature extraction and support vector machine (SVM) to achieve semi-automatic segmentation of IVOCT images. The image features utilized in this work including light attenuation coefficients and image textures based on gray level co-occurrence matrix. Different sets of hyper-parameters and image features were tested. This method achieved an accuracy of 83% on the test images. Single class accuracy of 89% for fibrous, 79.3% for calcification and 86.5% lipid tissue. The results show that this method can be a considerable way for semi-automatic segmentation of atherosclerotic plaque components in clinical IVOCT images.     

Validation of ThermoHuman automatic thermographic software for assessing foot temperature before and after running

The aim of the study was to evaluate an automatic thermographic software package (ThermoHuman®) for assessing skin temperature on the soles of the feet before and after running and to compare it with two manual definitions of the regions of interest (ROIs). 120 thermal images of the soles of the feet of 30 participants, at two measurement points (before and after running 30 min) and on two measurement days were analyzed. Three different models of thermographic image analyses were used to obtain the mean temperature of 9 ROIs: A) ThermoHuman (automatic definition of ROIs using ThermoHuman® software), B) Manual (manual delimitation of ROIs by proportion criteria), and C) Manual-TH (manual delimitation of ROIs in an attempt to replicate the regions analyzed by ThermoHuman). ThermoHuman resulted in an 86% reduction in time involved compared to manual delimitation. Fourteen of the 120 images (12%) presented some error in one or more of the ROI delimitations. Although the three procedures presented significant differences between them (53% in the comparison between ThermoHuman and Manual, 47% between ThermoHuman and Manual-TH, and 28% between Manual and Manual-TH), all differences had a small effect size (ES 0.2–0.4) or lower (ES < 0.2). Bland-Altman plots showed similar 95% limits of agreement between the three procedures before and after running. Intraclass correlation coefficient analysis of the three procedures presented excellent reliability (ICC>0.8). In conclusion, ThermoHuman® software was observed to be time-saving for image analysis with excellent reliability. Although results suggest that ThermoHuman® and manual methods are both valid in themselves, combining them is not recommended due to the differences observed between them.     

Convolution Neural Networks and Support Vector Machines for Automatic Segmentation of Intracoronary Optical Coherence Tomography

Cardiovascular diseases are closely associated with deteriorating atherosclerotic plaques. Optical coherence tomography (OCT) is a recently developed intravascular imaging technique with high resolution approximately 10 microns and could provide accurate quantification of coronary plaque morphology. However, tissue segmentation of OCT images in clinic is still mainly performed manually by physicians which is time consuming and subjective. To overcome these limitations, two automatic segmentation methods for intracoronary OCT image based on support vector machine (SVM) and convolutional neural network (CNN) were performed to identify the plaque region and characterize plaque components. In vivo IVUS and OCT coronary plaque data from 5 patients were acquired at Emory University with patient’s consent obtained. Seventy-seven matched IVUS and OCT slices with good image quality and lipid cores were selected for this study. Manual OCT segmentation was performed by experts using virtual histology IVUS as guidance, and used as gold standard in the automatic segmentations. The overall classification accuracy based on CNN method achieved 95.8%, and the accuracy based on SVM was 71.9%. The CNN-based segmentation method can better characterize plaque compositions on OCT images and greatly reduce the time spent by doctors in segmenting and identifying plaques.     

Mechanical analysis of arterial plaques in native geometry with OCT wall motion analysis

The mechanical behavior of an atherosclerotic plaque may encode information about the type, composition, and vulnerability to rupture. Human arterial segments with varying plaque burden were analyzed ex vivo with optical coherence tomography (OCT) to determine plaque type and to determine compliance during pulsatile inflation in their native geometry. Calcifications and lipid filled plaques showed markedly different compliance when analyzed with OCT wall motion analysis. There was also a trend towards increased circumferential variation in arterial compliance with increasing plaque burden.     

Three-dimensional spatial reconstruction of coronary arteries based on fusion of intravascular optical coherence tomography and coronary angiography

We present a three-dimensional (3D) spatial reconstruction of coronary arteries based on fusion of intravascular optical coherence tomography (IVOCT) and digital subtraction angiography (DSA). Centerline of vessel in DSA images is exacted by multi-scale filtering, adaptive segmentation, morphology thinning and Dijkstra's shortest path algorithm. We apply the cross-correction between lumen shapes of IVOCT and DSA images and match their stenosis positions to realize co-registration. By matching the location and tangent direction of the vessel centerline of DSA images and segmented lumen coordinates of IVOCT along pullback path, 3D spatial models of vessel lumen are reconstructed. Using 1121 distinct positions selected from eight vessels, the correlation coefficient between 3D IVOCT model and DSA image in measuring lumen radius is 0.94% and 97.7% of the positions fall within the limit of agreement by Bland–Altman analysis, which means that the 3D spatial reconstruction IVOCT models and DSA images have high matching level.     

Automated Detection Framework of the Calcified Plaque with Acoustic Shadowing in IVUS Images

Intravascular Ultrasound (IVUS) is one ultrasonic imaging technology to acquire vascular cross-sectional images for the visualization of the inner vessel structure. This technique has been widely used for the diagnosis and treatment of coronary artery diseases. The detection of the calcified plaque with acoustic shadowing in IVUS images plays a vital role in the quantitative analysis of atheromatous plaques. The conventional method of the calcium detection is manual drawing by the doctors. However, it is very time-consuming, and with high inter-observer and intra-observer variability between different doctors. Therefore, the computer-aided detection of the calcified plaque is highly desired. In this paper, an automated method is proposed to detect the calcified plaque with acoustic shadowing in IVUS images by the Rayleigh mixture model, the Markov random field, the graph searching method and the prior knowledge about the calcified plaque. The performance of our method was evaluated over 996 in-vivo IVUS images acquired from eight patients, and the detected calcified plaques are compared with manually detected calcified plaques by one cardiology doctor. The experimental results are quantitatively analyzed separately by three evaluation methods, the test of the sensitivity and specificity, the linear regression and the Bland-Altman analysis. The first method is used to evaluate the ability to distinguish between IVUS images with and without the calcified plaque, and the latter two methods can respectively measure the correlation and the agreement between our results and manual drawing results for locating the calcified plaque in the IVUS image. High sensitivity (94.68%) and specificity (95.82%), good correlation and agreement (>96.82% results fall within the 95% confidence interval in the Student t-test) demonstrate the effectiveness of the proposed method in the detection of the calcified plaque with acoustic shadowing in IVUS images.     

Asymmetric dimethylarginine and long-term adverse cardiovascular events in patients with type 2 diabetes: relation with the glycemic control

Patients with type 2 diabetes are at a high risk for acute cardiovascular events, which usually arise from the rupture of a vulnerable coronary lesion characterized by specific morphological plaque features. Thus, the identification of vulnerable plaques is of utmost clinical importance in patients with type 2 diabetes. However, there is currently no scoring system available to identify vulnerable lesions based on plaque characteristics. Thus, we aimed to characterize the diagnostic value of optical coherence tomography (OCT) - derived lesion characteristics to quantify plaque vulnerability both as individual parameters and when combined to a score in patients with type 2 diabetes. OCT was performed in the coronary culprit lesions of 112 patients with type 2 diabetes. The score, which quantifies plaque vulnerability, was defined as the predicted probability that a lesion is the cause for an acute coronary syndrome (ACS) (vs. stable angina (SAP)) based on its specific plaque morphology. Multivariable logistic regression analysis demonstrated that plaque vulnerability was independently predicted by the minimal fibrous cap thickness overlying a lesion’s lipid core (odds ratio (OR) per 10 μm 0.478, p = 0.002), the medium lipid arc (OR per 90° 13.997, p < 0.001), the presence of macrophages (OR 4.797, p = 0.015) and the lipid plaque length (OR 1.290, p = 0.098). Receiver-operating-characteristics (ROC) analyses demonstrated that these parameters combined to a score demonstrate an excellent diagnostic efficiency to identify culprit lesions of patients with ACS (vs. SAP, AUC 0.90, 95% CI 0.84-0.96). This is the first study to present a score to quantify lesion vulnerability in patients with type 2 diabetes. This score may be a valuable adjunct in decision-making and useful in guiding coronary interventions.     

Measurement of the transverse strain tensor in the coronary arterial wall from clinical intravascular ultrasound images

Atherosclerotic plaque rupture is the major cause of acute coronary syndromes. Currently, there is no reliable diagnostic tool to predict plaque rupture. Knowledge of plaque mechanical properties based on local artery wall strain measurements would be useful for characterizing its composition and predicting its vulnerability. Due to cardiac motion, strain estimation in clinical intravascular ultrasound (IVUS) images is extremely challenging. A method is presented to estimate cross-sectional coronary artery wall strain in response to cardiac pulsatile pressure using clinically acquired IVUS images, which are acquired in continuous pullback mode. First, cardiac phase information is retrieved retrospectively from an IVUS image sequence using an image-based gating method, and image sub-sequences at systole and diastole are extracted. Then, images at branch sites are used as landmarks to align the two image sub-sequences. Finally, the paired images at each site are registered to measure the 2D strain tensor of the coronary artery cross-section. This method has been successfully applied to IVUS images of a left anterior descending (LAD) coronary artery acquired clinically during a standard procedure. Such complete strain information should be useful for identifying vulnerable plaque.     

Markov random field based automatic image alignment for electron tomography

We present a method for automatic full-precision alignment of the images in a tomographic tilt series. Full-precision automatic alignment of cryo electron microscopy images has remained a difficult challenge to date, due to the limited electron dose and low image contrast. These facts lead to poor signal to noise ratio (SNR) in the images, which causes automatic feature trackers to generate errors, even with high contrast gold particles as fiducial features. To enable fully automatic alignment for full-precision reconstructions, we frame the problem probabilistically as finding the most likely particle tracks given a set of noisy images, using contextual information to make the solution more robust to the noise in each image. To solve this maximum likelihood problem, we use Markov Random Fields (MRF) to establish the correspondence of features in alignment and robust optimization for projection model estimation. The resulting algorithm, called Robust Alignment and Projection Estimation for Tomographic Reconstruction, or RAPTOR, has not needed any manual intervention for the difficult datasets we have tried, and has provided sub-pixel alignment that is as good as the manual approach by an expert user. We are able to automatically map complete and partial marker trajectories and thus obtain highly accurate image alignment. Our method has been applied to challenging cryo electron tomographic datasets with low SNR from intact bacterial cells, as well as several plastic section and X-ray datasets.     

Characterization of the human myocardium by optical coherence tomography

Imaging of cardiac tissue structure plays a critical role in the treatment and understanding of cardiovascular disease. Optical coherence tomography (OCT) offers the potential to provide valuable, high‐resolution imaging of cardiac tissue. However, there is a lack of comprehensive OCT imaging data of the human heart, which could improve identification of structural substrates underlying cardiac abnormalities. The objective of this study was to provide qualitative and quantitative analysis of OCT image features throughout the human heart. Fifty human hearts were acquired, and tissues from all chambers were imaged with OCT. Histology was obtained to verify tissue composition. Statistical differences between OCT image features corresponding to different tissue types and chambers were estimated using analysis of variance. OCT imaging provided features that were able to distinguish structures such as thickened collagen, as well as adipose tissue and fibrotic myocardium. Statistically significant differences were found between atria and ventricles in attenuation coefficient, and between adipose and all other tissue types. This study provides an overview of OCT image features throughout the human heart, which can be used for guiding future applications such as OCT‐integrated catheter‐based treatments or ex vivo investigation of structural substrates.     

Stress-Based Plaque Vulnerability Index and Assessment for Carotid Atherosclerotic Plaques Using Patient-Specific Vessel Material Properties

Cardiovascular diseases are closely linked to atherosclerotic plaque development and rupture. Assessment of plaque vulnerability is of fundamental significance to cardiovascular research and disease diagnosis, prevention, treatment and management. Magnetic resonance image (MRI) data of carotid atherosclerotic plaques from 8 patients (5 male, 3 female; age: 62-83, mean=71) were acquired at the University of Washington (UW), Seattle by the Vascular Imaging Laboratory (VIL) with written informed consent obtained. Patient-specific vessel material properties were quantified using Cine MRI data for modeling use. 3D thin-layer models were used to obtain plaque stress and strain for plaque assessment. A stress-based plaque vulnerability index (SPVI) was proposed to combine mechanical analysis, plaque morphology and composition for more complete carotid plaque vulnerability assessment. The five intervals (unit: kPa) [0, 46.8), [46.8, 80), [80, 92), [92, 103), and [103, +∞) from in vivo material models were used for SPVI values of 0, 1, 2, 3 and 4, respectively. The optimized agreement rate was 85.19%. The use of patient-specific material properties in plaque models could potentially improve the accuracy of model stress/strain calculations. SPVI has the potential to improve the current image-based screening and plaque vulnerability assessment schemes.     

In vivo detection of endotracheal tube biofilms in intubated critical care patients using catheter‐based optical coherence tomography

The formation of biofilms in the endotracheal tubes (ETTs) of intubated patients on mechanical ventilation is associated with a greater risk of ventilator‐associated pneumonia and death. New technologies are needed to detect and monitor ETTs in vivo for the presence of these biofilms. Longitudinal OCT imaging was performed in mechanically ventilated subjects at 24‐hour intervals until extubation to detect the formation and temporal changes of in vivo ETT biofilms. OCT‐derived attenuation coefficient images were used to differentiate between mucus and biofilm. Extubated ETTs were examined with optical and electron microscopy, and all imaging results were correlated with standard‐of‐care clinical test reports. OCT and attenuation coefficient images from four subjects were positive for ETT biofilms and were negative for two subjects. The processed and stained extubated ETTs and clinical reports confirmed the presence/absence of biofilms in all subjects. Our findings confirm that OCT can detect and differentiate between biofilm‐positive and biofilm‐negative groups (P < 10^?5). OCT image‐based features may serve as biomarkers for direct in vivo detection of ETT biofilms and help drive investigation of new management strategies to reduce the incidence of VAP.      

Repeatability Using Automatic Tracing with Canon OCT- HS100 and Zeiss Cirrus HD-OCT 5000

Background

Optical coherence tomography (OCT), can be used in clinical practice to provide high resolution cross-sectional images of the retina, optic disc and macula structure. These measurements can be useful for early detection, diagnosis, monitoring and treatment guidance for retinal diseases. Therefore, repeatability of measurements in OCT is of great importance.

Methods

Macula and optic disc parameters from the right eye of 30 healthy subjects were obtained twice with the Canon OCT-HS100 and Zeiss Cirrus HD-OCT 5000. Repeatability was evaluated by use of the coefficient of repeatability (CR) and the coefficient of repeatability as a percentage of the mean (CR%), and the obtained measurements were compared between the instruments.

Results

CR% of optic disc parameters ranged between 0.90 and 22.22% and 0.00 and 16.00% with the Canon and Zeiss OCT respectively. For macular parameters CR% ranged between 0.62 and 2.81% and 0.99 and 1.81% with the Canon and Zeiss OCT respectively. No statistical difference could be found when comparing the CR of all macular and disc measurements between the instruments. Compared to our previously published data repeatability has significantly improved with the inclusion of automatic tracking systems with both the Canon and Zeiss OCT.

Conclusion

Automatic tracking function improves repeatability in both Canon OCT-HS100 and Zeiss Cirrus HD-OCT 5000. However, measurements generated by the two instruments are still not interchangeable.     

Morphological and Stress Vulnerability Indices for Human Coronary Plaques and Their Correlations with Cap Thickness and Lipid Percent: An IVUS-Based Fluid-Structure Interaction Multi-patient Study

Plaque vulnerability, defined as the likelihood that a plaque would rupture, is difficult to quantify due to lack of in vivo plaque rupture data. Morphological and stress-based plaque vulnerability indices were introduced as alternatives to obtain quantitative vulnerability assessment. Correlations between these indices and key plaque features were investigated. In vivo intravascular ultrasound (IVUS) data were acquired from 14 patients and IVUS-based 3D fluid-structure interaction (FSI) coronary plaque models with cyclic bending were constructed to obtain plaque wall stress/strain and flow shear stress for analysis. For the 617 slices from the 14 patients, lipid percentage, min cap thickness, critical plaque wall stress (CPWS), strain (CPWSn) and flow shear stress (CFSS) were recorded, and cap index, lipid index and morphological index were assigned to each slice using methods consistent with American Heart Association (AHA) plaque classification schemes. A stress index was introduced based on CPWS. Linear Mixed-Effects (LME) models were used to analyze the correlations between the mechanical and morphological indices and key morphological factors associated with plaque rupture. Our results indicated that for all 617 slices, CPWS correlated with min cap thickness, cap index, morphological index with r = -0.6414, 0.7852, and 0.7411 respectively (p<0.0001). The correlation between CPWS and lipid percentage, lipid index were weaker (r = 0.2445, r = 0.2338, p<0.0001). Stress index correlated with cap index, lipid index, morphological index positively with r = 0.8185, 0.3067, and 0.7715, respectively, all with p<0.0001. For all 617 slices, the stress index has 66.77% agreement with morphological index. Morphological and stress indices may serve as quantitative plaque vulnerability assessment supported by their strong correlations with morphological features associated with plaque rupture. Differences between the two indices may lead to better plaque assessment schemes when both indices were jointly used with further validations from clinical studies.     

γH2AX foci as a measure of DNA damage: a computational approach to automatic analysis

The γH2AX focus assay represents a fast and sensitive approach for the detection of one of the critical types of DNA damage - double-strand breaks (DSB) induced by various cytotoxic agents including ionising radiation. Apart from research applications, the assay has a potential in clinical medicine/pathology, such as assessment of individual radiosensitivity, response to cancer therapies, as well as in biodosimetry. Given that generally there is a direct relationship between numbers of microscopically visualised γH2AX foci and DNA DSB in a cell, the number of foci per nucleus represents the most efficient and informative parameter of the assay. Although computational approaches have been developed for automatic focus counting, the tedious and time consuming manual focus counting still remains the most reliable way due to limitations of computational approaches. We suggest a computational approach and associated software for automatic focus counting that minimises these limitations. Our approach, while using standard image processing algorithms, maximises the automation of identification of nuclei/cells in complex images, offers an efficient way to optimise parameters used in the image analysis and counting procedures, optionally invokes additional procedures to deal with variations in intensity of the signal and background in individual images, and provides automatic batch processing of a series of images. We report results of validation studies that demonstrated correlation of manual focus counting with results obtained using our computational algorithm for mouse jejunum touch prints, mouse tongue sections and human blood lymphocytes as well as radiation dose response of γH2AX focus induction for these biological specimens.     

Automatic lumen segmentation using uniqueness of vascular connected region for intravascular optical coherence tomography

We present an automatic lumen segmentation method using uniqueness of connected region for intravascular optical coherence tomography (IVOCT), which can effectively remove the effect on lumen segmentation caused by blood artifacts. Utilizing the uniqueness of vascular wall on A-lines, we detect the A-lines shared by multiple connected regions, identify connected regions generated by blood artifacts using traversal comparison of connected regions' location, shared ratio and area ratio and then remove all artifacts. We compare these three methods by 216 challenging images with severe blood artifacts selected from clinical 1076 IVOCT images. The metrics of the proposed method are evaluated including Dice index, Jaccard index and accuracy of 94.57%, 90.12%, 98.02%. Compared with automatic lumen segmentation based on the previous morphological feature method and widely used dynamic programming method, the metrics of the proposed method are significantly enhanced, especially in challenging images with severe blood artifacts.     

通过光学相干断层扫描图像进行自动视网膜分割评估黄斑水肿的投影面积

在本文中,我们提出了一种自动视网膜分割方法,以评估光学相干断层扫描（OCT）图像中黄斑水肿（ME）在视网膜特定层上的投影面积。首先使用基于权重矩阵的优化的最短路径最快算法对十个视网膜层边界进行分割,这有效降低了算法对血管阴影的敏感性。然而,ME的存在将导致水肿区域的分割不准确。因此,我们使用强度阈值方法提取每个OCT图像中的水肿区域,并将该区域中的值设置为零,并确保获得的分割边界可以自动穿过而不是绕过水肿区域。我们使用最小值投影来计算ME在不同层的投影面积。为了测试我们的方法,我们使用了从Topcon的OCT机器收集的数据。在轴向和B扫描方向上测得的黄斑区域分辨率分别为11.7微米和46.8微米。与手动分割相比,视网膜层边界分割的平均绝对误差和标准偏差为4.5±3.2微米。因此,所提出的方法为评估水肿提供了一种自动,无创和定量的工具。     

Cyclic Bending Contributes to High Stress in a Human Coronary Atherosclerotic Plaque and Rupture Risk: In Vitro Experimental Modeling and Ex Vivo MRI-Based Computational Modeling Approach

Many acute cardiovascular syndromes such as heart attack and stroke are caused by atherosclerotic plaque ruptures which often happen without warning. MRI-based models with fluid-structure interactions (FSI) have been introduced to perform flow and stress/strain analysis for atherosclerotic plaques and identify possible mechanical and morphological indices for accurate plaque vulnerability assessment. In this paper, cyclic bending was added to 3D FSI coronary plaque models for more accurate mechanical predictions. Curvature variation was prescribed using the data of a human left anterior descending (LAD) coronary artery. Five computational models were constructed based on ex vivo MRI human coronary plaque data to assess the effects of cyclic bending, pulsating pressure, plaque structure, and axial stretch on plaque stress/strain distributions. In vitro experiments using a hydrogel stenosis model with cyclical bending were performed to observe effect of cyclical bending on flow conditions. Our results indicate that cyclical bending may cause more than 100% or even up to more than 1000% increase in maximum principal stress values at locations where the plaque is bent most. Stress increase is higher when bending is coupled with axial stretch, non-smooth plaque structure, or resonant pressure conditions (zero phase angle shift). Effects of cyclic bending on flow behaviors are more modest (21.6% decrease in maximum velocity, 10.8% decrease in flow rate, maximum flow shear stress changes were < 5%). Computational FSI models including cyclic bending, plaque components and structure, axial stretch, accurate in vivo measurements of pressure, curvature, and material properties should lead to significant improvement on stress-based plaque mechanical analysis and more accurate coronary plaque vulnerability assessment.     

Segmentation of multi-isotope imaging mass spectrometry data for semi-automatic detection of regions of interest

Multi-isotope imaging mass spectrometry (MIMS) associates secondary ion mass spectrometry (SIMS) with detection of several atomic masses, the use of stable isotopes as labels, and affiliated quantitative image-analysis software. By associating image and measure, MIMS allows one to obtain quantitative information about biological processes in sub-cellular domains. MIMS can be applied to a wide range of biomedical problems, in particular metabolism and cell fate [1], [2], [3]. In order to obtain morphologically pertinent data from MIMS images, we have to define regions of interest (ROIs). ROIs are drawn by hand, a tedious and time-consuming process. We have developed and successfully applied a support vector machine (SVM) for segmentation of MIMS images that allows fast, semi-automatic boundary detection of regions of interests. Using the SVM, high-quality ROIs (as compared to an expert's manual delineation) were obtained for 2 types of images derived from unrelated data sets. This automation simplifies, accelerates and improves the post-processing analysis of MIMS images. This approach has been integrated into "Open MIMS," an ImageJ-plugin for comprehensive analysis of MIMS images that is available online at http://www.nrims.hms.harvard.edu/NRIMS_ImageJ.php.     

Shear‐induced diffusion of red blood cells measured with dynamic light scattering‐optical coherence tomography

Quantitative measurements of intravascular microscopic dynamics, such as absolute blood flow velocity, shear stress and the diffusion coefficient of red blood cells (RBCs), are fundamental in understanding the blood flow behavior within the microcirculation, and for understanding why diffuse correlation spectroscopy (DCS) measurements of blood flow are dominantly sensitive to the diffusive motion of RBCs. Dynamic light scattering‐optical coherence tomography (DLS‐OCT) takes the advantages of using DLS to measure particle flow and diffusion within an OCT resolution‐constrained three‐dimensional volume, enabling the simultaneous measurements of absolute RBC velocity and diffusion coefficient with high spatial resolution. In this work, we applied DLS‐OCT to measure both RBC velocity and the shear‐induced diffusion coefficient within penetrating venules of the somatosensory cortex of anesthetized mice. Blood flow laminar profile measurements indicate a blunted laminar flow profile and the degree of blunting decreases with increasing vessel diameter. The measured shear‐induced diffusion coefficient was proportional to the flow shear rate with a magnitude of ~0.1 to 0.5 × 10^?6 mm². These results provide important experimental support for the recent theoretical explanation for why DCS is dominantly sensitive to RBC diffusive motion.